We listened to your requests and feedback for new and improved features and have made significant improvements across all the products in our market leading AI Platform. In 6.1 we focus on business value with a new Use Case Value Tracker, Profit Curves, and Payoff Matrices. We have also introduced spatial-aware modeling that we are calling Location AI. Anomaly Detection for Time Series is new, and we have also introduced the concept of Champion/Challenger Models for MLOps. In our UI, we even have a new light mode theme!
Here’s a highlight of the major components delivered in Release 6.1. (Make sure to tell us what you think of this material.)
Use Case Value Tracker. We are constantly adding features to help with the business operationalization of your machine learning efforts. The new Use Case Value Tracker is a centralized hub for collaborating with team members around a single AI initiative, end-to-end. Use Case Value Tracker enables you to manage your machine learning projects and understand the associated value at every step. You can organize all your DataRobot assets around a given use case. For example, you can group all of the datasets, models, deployments, and apps associated with a customer acquisition initiative. You can also see metrics on the realized business value of your use case over time. This helps you understand the true ROI of your enterprise AI.
Location AI. We recently secured a provisional patent and added spatial awareness to our Automated Machine Learning product. Location AI allows your predictive models to understand spatial relationships between observations in a dataset. For example, one of the best ways to predict home prices is to look at the prices of other homes in the same neighborhood. Knowing the effects of proximity is critical for many prediction problems and Location AI automates complex spatial modeling for both novice and expert users. You can even mix your location data with other data types including tabular, text, and images.
Profit Curve and Payoff Matrix. Most organizations who are serious about AI need to be able to assess the financial impact of the predictions their models generate. In DataRobot 6.1 we introduce Profit Curve and Payoff Matrix so you can visualize the payoff from good predictions and the costs associated with bad predictions. You can set up a number of alternative payoff matrices, inspect the resulting profit curve, and even compare alternative profit curves on the same chart. Armed with these insights you can decide the most profitable option to take. DataRobot's Profit Curve and Payoff Matrices help you tune your models for business impact not just accuracy.
MLOps Challenger Models. Accurate models today may not be accurate tomorrow, and Champion/Challenger models in MLOps Release 6.1 enable you to test and compare your production models with alternatives to determine the most appropriate and accurate as business conditions change. This allows you to continue to perform the same rigor of analysis on your production models as you did during training. You can replay predictions against challenger models to analyze accuracy and performance over the same time period. When a challenger beats out the current champion, you can instantly switch out your old model with absolutely no service interruption.
Anomaly Detection for Automated Time Series. In this release we introduce Automated Time Series Anomaly Detection. Without the need to specify a target variable, you can now train time series models in a fully unsupervised mode. DataRobot automatically selects, builds, tests, and ranks a diverse set of anomaly detection models and uses a novel Synthetic AUC error metric to help you understand which model is best for your specific use case. DataRobot's Automated Time Series Anomaly Detection also provides all of the automated insights and visualizations to help you explain every model and drill into individual anomalies to understand causal factors. Anomaly Detection unlocks a wide variety of new use cases for Automated Time Series customers. For example, you could build a model to monitor pressure sensors on a water pump. This helps you can perform preventative maintenance when readings are abnormally high thereby avoiding expensive pump failures.
More Explainability and Trust Features. With every release we aim to bring more features to help you explain and trust your AI. This release is no exception. We have introduced an automated Data Quality Assessment and Data Quality Handling Report that surfaces and reports on issues such as missing values, outliers, target leakage, and more. We now provide intuitive SHAP explanations to show you exactly how much each feature is responsible for predictions that deviate from the average, and a unique Humble AI feature in MLOps that allows you to define rules and take corrective action when a production model makes an uncertain prediction e.g., override prediction to 'loan is bad' when loan amount is greater than $100,000.
Other New Features and Enhancements. Release 6.1 is jam-packed with new capabilities that you’ll want to take advantage of immediately, including Spark SQL data prep and other enhancements to Paxata, Visual AI (now enabled by default) with Prediction Explanations, tight integration with popular third-party products including Snowflake, MS SQL Server, and Tableau and dozens of usability improvements including a new UI light mode theme.
These new capabilities support our ongoing commitment to deliver the most powerful platform for end-to-end enterprise AI.
Public Beta features
Every DataRobot release includes some features that have been tested by our engineering and quality teams, and are available for preview by a limited number of users. If you're a customer and are interested in giving some of these features a try, contact your CFDS or account executive for details on how you can participate.
Here are some public beta features you can check out now:
AI Catalog enhancements—Bulk Operations
Feature Discovery enhancements—relationship editor, prediction points, feature reduction, etc.
Time Series enhancements—new Forecast vs. Actuals Plot, Anomaly Assessment Insights
We’ve created a series of blogs and demo videos to help you understand the changes and guide you while using this release. New blogs will be published periodically during the next two weeks so please make sure to check our blog portal often. For the best experience, you can also subscribe to our blog and we’ll notify you when new blogs are posted.
Below are demo videos for many of the new and updated features.
DataRobot UI Themes (copy and share link to this demo)
Data Access/Preparation & Feature Engineering
Data Preparation Using Spark SQL (copy and share link to this demo)
Automated Feature Discovery—New Workflows and Relationship Editor (PUBLIC BETA) (copy and share link to this demo)
Dataset and Model Comments (copy and share link to this demo)
Modeling and Evaluation
New Rerun Autopilot Options (copy and share link to this demo)
Core Modeling—Profit Curve and Payoff Matrix (copy and share link to this demo)
Explainable AI—Feature Impact Custom Sample Size (copy and share link to this demo)
Explainable AI—Data Quality Assessment (copy and share link to this demo)
Explainable AI—SHAP Blueprints and Prediction Explanations (copy and share link to this demo)
Visual AI—Image Prediction Explanations (copy and share link to this demo)
Visual AI—Data Quality Assessment (copy and share link to this demo)
Time Series—Anomaly Detection (copy and share link to this demo)
Unsupervised Modeling UI in AutoML (copy and share link to this demo)
Location AI (Public Beta feature)
Derived Geometric Features (copy and share link to this demo)
Native File Support (copy and share link to this demo)
Spatial Featurizers (copy and share link to this demo)
Accuracy Over Space (copy and share link to this demo)
Map Visualizations in EDA (copy and share link to this demo)
Auto-recognition of Location Variables (copy and share link to this demo)
Scheduled Integration Jobs (copy and share link to this demo)
DataRobot Snowflake Integration (copy and share link to this demo)
DataRobot Tableau Integration (copy and share link to this demo)
Google BigQuery as a Data Source (copy and share link to this demo)
Monitoring Agents (copy and share link to this demo)
More information for DataRobot users: search in-app Platform Documentation for Release Center, and find the Version 6.1.0 release notes for AutoML or MLOps.
Tell us what you think
If this material is helpful, let us know. If you still have questions about the release, let us know. Other DataRobot Community members and our DataRobot experts will help fill in any blanks, and your feedback is critical to helping us get better. Click Comment and let us know.
We're excited to welcome Zepl and its employees and customers to DataRobot! The acquisition of Zepl and integration of its self-service data science notebook solution will provide additional flexibility for data scientists who prefer to code. In her blog article, Tricia Lee explains how you can check out Zepl today. Please have a look, give it a spin, and let us know what you think.